The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes npm run build to verify that visual modifications do not break the project's build process. This is a standard and expected operation for a code-refactoring skill.
[PROMPT_INJECTION]: The skill instructions are focused on task performance and do not contain patterns intended to bypass safety protocols. While the skill reads untrusted project files (creating an indirect prompt injection surface), this behavior is essential for its primary function.
[DATA_EXFILTRATION]: No network operations or patterns associated with data exfiltration were found. Data access is restricted to the local game project context.
[EXTERNAL_DOWNLOADS]: The skill does not perform any remote downloads or execution of external scripts.

design-game